Skip to main content
ABC News
Can You Buy A Presidential Election?

After spending $1 billion on a presidential run, former New York Mayor Michael Bloomberg dropped out of the race for the Democratic nomination following Super Tuesday, securing just 44 pledged delegates. The national conversation quickly moved on, but the results of Bloomberg’s experiment are worth examining in more detail.

His campaign wasn’t successful, but did his spending have any impact at all?

My colleagues and I set out to answer this question by linking data on Bloomberg’s TV ad spending1 with rolling national survey data from the Democracy Fund + UCLA Nationscape, and based on the results, we can infer that Bloomberg’s advertising did increase his support in media markets where he advertised. Ultimately, however, his ads weren’t enough to overcome the challenges of rising press scrutiny and poor debate showings.

In media markets where Bloomberg advertised, he received 10 percent support, on average, in the Nationscape survey of the Democratic primary across the 12-week period from mid-December 2019 to late February 2020, compared with 4 percent in markets where he did not advertise.2 And this effect holds true even when we control for the demographics of survey respondents and the primary calendar.3

These numbers might not sound particularly impressive, especially considering Bloomberg spent $1 billion overall, but as you can see in the chart below, his advertising push did pay off. We found Bloomberg’s support grew more than three times as fast in markets where he advertised than where he didn’t, and in the week before his disastrous debate performance on Feb. 19, he registered 15 percent support in these markets compared with 6 percent in other markets.

Of course, that Feb. 19 debate changed everything for Bloomberg.

It was his first time taking the debate stage, but his performance was widely regarded as poor. He took criticism for backing stop-and-frisk policing tactics as mayor and refused to address allegations of sexism and fostering a hostile workplace for women at his company. His national support in the Nationscape survey fell 1 percentage point the week after the debate and 1 point the week before Super Tuesday.

Liberal voter support in markets where he’d advertised was cut in half after the debate as well, from 11 percent the week before the debate to 9 percent after, and just 6 percent the week before Super Tuesday. Support among Democrats age 50 and over in these markets — the age demographic most likely to be watching TV — fell, too, from 21 to 18 percent before dipping slightly to 17 percent in the last week of his campaign.

The logic of Bloomberg’s media blitz was that it would generate enough support in the Super Tuesday states for him to easily clear states’ 15 percent pledged delegate thresholds and allow him to command a sizable share of delegates in a fragmented field. But when Bloomberg ran into trouble after the debate, his support among those in markets with his advertising proved soft and more easily reversible.4

Bloomberg’s support from TV advertising also seemed to have its limits. For instance, we found a minimal effect on his Nationscape polling numbers in markets where he spent more money. Grouping markets on a per-household measure of spending, Bloomberg’s support is just 1 point higher in markets where he spent more (10 percent) than in markets where he spent less (9 percent).5

On its own, Bloomberg’s experiment shows TV advertising can’t swing an entire election. It could not overcome Bloomberg’s lack of charisma or skill as a debater. Bloomberg’s campaign did show, however, that advertising can have a measurable, double-digit impact on the polls and vault a candidate into the top tier. That’s not nothing.

FiveThirtyEight Politics Podcast: The data behind police violence


  1. My firm Echelon Insights built a database of political TV advertising buys sourced from the Federal Communications Commission, which we used to link local Bloomberg ad buys to the geographic location of survey respondents at the media market level.

  2. Bloomberg did advertise nationally — during the Super Bowl, for example, and online — but the majority of his advertising budget went toward local TV ad buys, so exposure to his ads was much greater in local markets where he advertised.

  3. Bloomberg did not campaign or heavily advertise in the first four states to vote (Iowa, New Hampshire, Nevada and South Carolina), but survey respondents in those states could still indicate their preference for him as the Democratic nominee. That said, his 4 percent support in the early states matched his support in states that voted later where he didn’t advertise.

  4. Bloomberg advertised in areas with more than 85 percent of the Democratic primary respondents in the survey, so we would expect the weekly trend in markets where he didn’t advertise to be noisier. However, we didn’t find an apparent decline in Bloomberg support in those areas after the debate. In fact, combining responses from the two-week periods before and after the debate shows a slight uptick in support.

  5. We excluded New York City from this particular analysis, as Bloomberg’s support was higher there as a former mayor.

Patrick Ruffini worked as a digital strategist for George W. Bush’s 2004 campaign, and the Republican National Committee in 2006. He co-founded Echelon Insights and founded Engage.